A self-decomposable distribution is a statistical distribution such that, for every ,
a random variable
with that distribution can be represented in distribution
as
, where
has the same distribution as
and is independent of
(Schilling et al. 2012). Every self-decomposable
distribution is an infinitely divisible
distribution.
For a strictly positive probability density function on
,
Mao and Tian (2026) proved that logarithmic convexity together with geometric concavity
is sufficient for self-decomposability. Here logarithmic convexity means that
is convex, and geometric concavity means that
is concave. For
differentiable
,
the latter condition is equivalent to
being nonincreasing. The conditions are sufficient
and need not describe the entire parameter range of self-decomposable laws. The same
authors proved that if
is logarithmically convex and
is nondecreasing, then a random
variable with density
has the distribution of
, where
has density
and
is independent of
.
For the Whittaker density proportional to on
, where
,
these criteria prove self-decomposability when
and
.
They also give an independent nonnegative additive remainder when
and
are increased by arbitrary nonnegative amounts from a starting
pair with
and
(Mao and Tian 2026).